block/qcow2.h | 4 ++ include/block/block.h | 8 ++- block/io.c | 9 ++- block/qcow2-refcount.c | 32 ++++++++++ block/qcow2.c | 11 ++++ tests/perf/block/qcow2/convert-blockstatus | 71 ++++++++++++++++++++++ tests/qemu-iotests/102 | 2 +- tests/qemu-iotests/102.out | 3 +- tests/qemu-iotests/141.out | 2 +- tests/qemu-iotests/144.out | 2 +- 10 files changed, 138 insertions(+), 6 deletions(-) create mode 100755 tests/perf/block/qcow2/convert-blockstatus
Hi!
It's a continuation for
"[PATCH] qcow2: avoid lseek on block_status if possible"
https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06598.html
performance results for block-status on tmpfs [tests originally by Kevin,
now they are in 01]:
./tests/perf/block/qcow2/convert-blockstatus /ramdisk/x
after 01:
plain: 81.77
forward: 82.61
prealloc: 0.01
after 02:
plain: 0.12
forward: 0.12
prealloc: 0.01
v2:
01: new
02: [mostly by Kevin's comments]
- rewritten to go through new flag BDRV_BLOCK_RECURSE
- never retry detection if failed for first time
- rewrite detection to do less iterations and to be more simple
iotests 102 behavior changed [and not sure about other two ones
in comparison with v1, but it seems it doesn't matter]
also, patch subject changed, as now it's a generic change for
block layer
Vladimir Sementsov-Ogievskiy (2):
tests/perf: Test lseek influence on qcow2 block-status
block: avoid recursive block_status call if possible
block/qcow2.h | 4 ++
include/block/block.h | 8 ++-
block/io.c | 9 ++-
block/qcow2-refcount.c | 32 ++++++++++
block/qcow2.c | 11 ++++
tests/perf/block/qcow2/convert-blockstatus | 71 ++++++++++++++++++++++
tests/qemu-iotests/102 | 2 +-
tests/qemu-iotests/102.out | 3 +-
tests/qemu-iotests/141.out | 2 +-
tests/qemu-iotests/144.out | 2 +-
10 files changed, 138 insertions(+), 6 deletions(-)
create mode 100755 tests/perf/block/qcow2/convert-blockstatus
--
2.18.0
ping2 Hi! My approach of adding perf test disliked by Max in other thread, so 01 may be just skipped. What about 02? 08.04.2019 19:26, Vladimir Sementsov-Ogievskiy wrote: > Hi! > > It's a continuation for > "[PATCH] qcow2: avoid lseek on block_status if possible" > https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06598.html > > performance results for block-status on tmpfs [tests originally by Kevin, > now they are in 01]: > > ./tests/perf/block/qcow2/convert-blockstatus /ramdisk/x > > after 01: > > plain: 81.77 > forward: 82.61 > prealloc: 0.01 > > after 02: > > plain: 0.12 > forward: 0.12 > prealloc: 0.01 > > v2: > 01: new > 02: [mostly by Kevin's comments] > - rewritten to go through new flag BDRV_BLOCK_RECURSE > - never retry detection if failed for first time > - rewrite detection to do less iterations and to be more simple > > iotests 102 behavior changed [and not sure about other two ones > in comparison with v1, but it seems it doesn't matter] > > also, patch subject changed, as now it's a generic change for > block layer > > > Vladimir Sementsov-Ogievskiy (2): > tests/perf: Test lseek influence on qcow2 block-status > block: avoid recursive block_status call if possible > > block/qcow2.h | 4 ++ > include/block/block.h | 8 ++- > block/io.c | 9 ++- > block/qcow2-refcount.c | 32 ++++++++++ > block/qcow2.c | 11 ++++ > tests/perf/block/qcow2/convert-blockstatus | 71 ++++++++++++++++++++++ > tests/qemu-iotests/102 | 2 +- > tests/qemu-iotests/102.out | 3 +- > tests/qemu-iotests/141.out | 2 +- > tests/qemu-iotests/144.out | 2 +- > 10 files changed, 138 insertions(+), 6 deletions(-) > create mode 100755 tests/perf/block/qcow2/convert-blockstatus > -- Best regards, Vladimir
Am 08.04.2019 um 18:26 hat Vladimir Sementsov-Ogievskiy geschrieben: > Hi! > > It's a continuation for > "[PATCH] qcow2: avoid lseek on block_status if possible" > https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06598.html > > performance results for block-status on tmpfs [tests originally by Kevin, > now they are in 01]: > > ./tests/perf/block/qcow2/convert-blockstatus /ramdisk/x > > after 01: > > plain: 81.77 > forward: 82.61 > prealloc: 0.01 > > after 02: > > plain: 0.12 > forward: 0.12 > prealloc: 0.01 Thanks, applied to the block branch. Kevin
22.05.2019 13:47, Kevin Wolf wrote: > Am 08.04.2019 um 18:26 hat Vladimir Sementsov-Ogievskiy geschrieben: >> Hi! >> >> It's a continuation for >> "[PATCH] qcow2: avoid lseek on block_status if possible" >> https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06598.html >> >> performance results for block-status on tmpfs [tests originally by Kevin, >> now they are in 01]: >> >> ./tests/perf/block/qcow2/convert-blockstatus /ramdisk/x >> >> after 01: >> >> plain: 81.77 >> forward: 82.61 >> prealloc: 0.01 >> >> after 02: >> >> plain: 0.12 >> forward: 0.12 >> prealloc: 0.01 > > Thanks, applied to the block branch. > > Kevin > Thank you! -- Best regards, Vladimir
ping 08.04.2019 19:26, Vladimir Sementsov-Ogievskiy wrote: > Hi! > > It's a continuation for > "[PATCH] qcow2: avoid lseek on block_status if possible" > https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06598.html > > performance results for block-status on tmpfs [tests originally by Kevin, > now they are in 01]: > > ./tests/perf/block/qcow2/convert-blockstatus /ramdisk/x > > after 01: > > plain: 81.77 > forward: 82.61 > prealloc: 0.01 > > after 02: > > plain: 0.12 > forward: 0.12 > prealloc: 0.01 > > v2: > 01: new > 02: [mostly by Kevin's comments] > - rewritten to go through new flag BDRV_BLOCK_RECURSE > - never retry detection if failed for first time > - rewrite detection to do less iterations and to be more simple > > iotests 102 behavior changed [and not sure about other two ones > in comparison with v1, but it seems it doesn't matter] > > also, patch subject changed, as now it's a generic change for > block layer > > > Vladimir Sementsov-Ogievskiy (2): > tests/perf: Test lseek influence on qcow2 block-status > block: avoid recursive block_status call if possible > > block/qcow2.h | 4 ++ > include/block/block.h | 8 ++- > block/io.c | 9 ++- > block/qcow2-refcount.c | 32 ++++++++++ > block/qcow2.c | 11 ++++ > tests/perf/block/qcow2/convert-blockstatus | 71 ++++++++++++++++++++++ > tests/qemu-iotests/102 | 2 +- > tests/qemu-iotests/102.out | 3 +- > tests/qemu-iotests/141.out | 2 +- > tests/qemu-iotests/144.out | 2 +- > 10 files changed, 138 insertions(+), 6 deletions(-) > create mode 100755 tests/perf/block/qcow2/convert-blockstatus > -- Best regards, Vladimir
ping 08.04.2019 19:26, Vladimir Sementsov-Ogievskiy wrote: > Hi! > > It's a continuation for > "[PATCH] qcow2: avoid lseek on block_status if possible" > https://lists.gnu.org/archive/html/qemu-devel/2019-01/msg06598.html > > performance results for block-status on tmpfs [tests originally by Kevin, > now they are in 01]: > > ./tests/perf/block/qcow2/convert-blockstatus /ramdisk/x > > after 01: > > plain: 81.77 > forward: 82.61 > prealloc: 0.01 > > after 02: > > plain: 0.12 > forward: 0.12 > prealloc: 0.01 > > v2: > 01: new > 02: [mostly by Kevin's comments] > - rewritten to go through new flag BDRV_BLOCK_RECURSE > - never retry detection if failed for first time > - rewrite detection to do less iterations and to be more simple > > iotests 102 behavior changed [and not sure about other two ones > in comparison with v1, but it seems it doesn't matter] > > also, patch subject changed, as now it's a generic change for > block layer > > > Vladimir Sementsov-Ogievskiy (2): > tests/perf: Test lseek influence on qcow2 block-status > block: avoid recursive block_status call if possible > > block/qcow2.h | 4 ++ > include/block/block.h | 8 ++- > block/io.c | 9 ++- > block/qcow2-refcount.c | 32 ++++++++++ > block/qcow2.c | 11 ++++ > tests/perf/block/qcow2/convert-blockstatus | 71 ++++++++++++++++++++++ > tests/qemu-iotests/102 | 2 +- > tests/qemu-iotests/102.out | 3 +- > tests/qemu-iotests/141.out | 2 +- > tests/qemu-iotests/144.out | 2 +- > 10 files changed, 138 insertions(+), 6 deletions(-) > create mode 100755 tests/perf/block/qcow2/convert-blockstatus > -- Best regards, Vladimir
© 2016 - 2024 Red Hat, Inc.